The State of Link Building 2016: What I Learned Manually Analysing 1,000 Search Results

HOW TO EARN COMMISSION FORM CLICKBANK ON AUTO-PILOT FOR LIFETIME
Click Banner

168 Comments 18 minutes


analysisDo private blog network’s still work? Does a higher word-count help your pages rank better? Did Glen really spend 60 hours on this article? I hope to answer all of these questions and many more in my new behind the scenes report on the current state of link building.

I can clarify I did spend more than 60 hours of work on this article, yet the sad part for me is that most of that time can be summarized in a simple bar chart. The rest of the time was spent coming up with a good headline but I clearly failed at that, so let’s see if I did any better with the chart.

I Manually Analysed 1,000 Search Results to See How Websites Ranked

I’ve been guilty over the years of making generalizations like “private blog networks are dominating Google” or “natural link building is almost impossible in some industries” so a few weeks ago I decided that I would respond to my own sweeping statements and analyze how people are actually ranking their websites in 2016.

As you can imagine, doing this analysis manually was a very time-consuming process. I managed to overcome most of the monotony by seeing this work as a chance to discover more link opportunities for myself (and my clients). My private database grew by over a hundred rows which means that there were many replicable links in my findings.

Now, before the SEO world tells me how unscientific the following data is, allow me this one caveat: I agree. The following findings are primarily based on my personal experience and viewpoints. There is, unfortunately, no way to exactly determine which backlinks are most integral in helping a web page rank.

The Results

The goal of my research was simple: Which specific type of link was the most instrumental in helping a website to rank.

Of course, every website I reviewed of course received backlinks from a number of different sources but I wanted to discover which ones were helping that particular website the most.

Because this was performed manually – I couldn’t automate the process even if I wanted to – I understand that there is nothing exact with my findings.

There are said to be over 200 factors which Google use to rank websites and while links from other websites are certainly the most impactful, it’s possible that my personal views are not entirely what is helping these sites rise to the top of search results.

That being said, I’ve been doing SEO for 11 years now and much of that time has been spent on link building. I wanted these answers for myself, so there is hopefully some merit in the following data.

Enough with the writing. Here are the results.

prominent

That’s it. The equivalent of working two and a half days straight without taking even a one-second break mostly boils down to that single graph.

As you can see, what I consider to be ‘natural’ link building tops the chart. This really shouldn’t be too surprising since that is how Google is supposed to rank websites (for the most part).

I should add that I don’t believe 21% of these results I checked were ranking because of links. Some were on powerful domains like Youtube.com or Amazon.com and therefore were ranking primarily because of the domain the result resided on. These links were still analysed, with most coming under the mixed category.

Due to the industries I analysed (revealed further down) there’s also a chance that there are more ‘low-quality’ links then you would find with a much broader dataset. However, you’ll find I picked the terms I monitored for good reason.

There were two key things that surprised me with these results:

  • How low in quality the backlinks were to many top ranking websites
  • How few private link networks I uncovered

The second point was especially interesting to me as it feels like I’m finding private link networks on a daily basis. What’s probably happening is that my brain makes some kind of internal ‘event’ when I come across one and therefore I’m less likely to remember all of the times I didn’t find them.

Kind of like how when you purchase a new car you start suddenly seeing it everywhere yet you didn’t even notice it before.

More seasoned SEO’s will probably be interested in how I classified those links but for the most part, by ‘low-quality link’s’ I generally mean links that anyone can replicate with a high level of ease and they weren’t earned in any way.

Link Classifications

The six categories that I have chosen to split links up into are:

  • Natural
  • Press Releases / Articles
  • Poor Links / Spam
  • Mixed Links
  • Network Links
  • Guest Blogging Links

To clarify again that my decisions are based on what I believe the strongest links the site has are.

Natural

By natural I simply mean that while a webmaster may have a mix of links, they are earned links rather than those that appear to have been gained in order to increase search engine rankings.

Though SEO may be a consideration at times – such as utilising signatures in forum posts – they’re essentially the types of links that you would happily show a Google reviewer and not be concerned about.

Press Releases / Articles

Sites in this category derive their rankings primarily from using press release services which allow you to embed links or embedding them in article directories which allow you to post your own content.

Low-Quality Links

These are primarily links that people can build either manually or automatically with tools that were likely built just to influence search results.

The types of links here include things like irrelevant blog comments, forum profile pages, social bookmarking links and very often from non-English Blogspot blogs.

Mixed Links

Sites in ‘mixed links’ appeared to have a bit of every type of link without any certain type – at least to me – being a major factor in why the site was ranking.

Though not all here used guest blogging or network links, mixed means that they had some natural links and some that were clearly built for gaming Google.

Network Links

This is for sites whose rankings clearly rely on the ownership of a private link network (often known as a private blog network, or PBN). While I can’t be certain sites were utilising their own PBN, it’s highly unlikely an outside source did it as a form of negative SEO, and – let’s be fair – it’s very easy to tell what’s going on when you find a network.

Guest Blogging Links

Though many webmasters did utilising guest blogging, few seemed to benefit from it as their main source of links. In fact, I only found a handful of webmasters primarily benefiting from this.

I’ve Already Got the Data, What Else Can We Analyse?

Since I was already relegated to the idea that I was going to analyse all of these search results anyway, I decided that I may as well collect more data on the way in the hope it would produce some more interesting charts.

Once again I’ll be the first to admit that this is far from scientific. Brian has a much better analysis with 1 million search results if you want some broader results. My sample size is admittedly too small to set the SEO world on fire with the following graphs but I still thought it would be interesting to analyse.

In the GIF below so you can see that all of this data really was collected manually. Huge thanks to my brother who I roped in to help with the grunt work on this.

data-gif

Where I have tried to separate myself from the likes of Brian’s data is that I’ve specifically monitored industries that you could make money in if you were to rank on the first page of Google.

With Brian’s data, I have no idea if those million search queries were focused on the medical field or other technical subjects which simply wouldn’t apply to what the majority of us are trying to rank for in Google.

The Clickbank affiliate marketplace was a big inspiration for my keyword choices since people are successfully selling products in the industries I monitored. Here’s a sample of the keywords that I analysed:

example

I am aware, as stated above, my search queries of choice would likely result in more lower-quality link profiles than the web as a whole but again, I wanted to look at industries that myself and ViperChill readers are more likely looking to rank in.

Number of Backlinks

We all know that backlinks aren’t created equal, but would the data support that?

backlinks

I can see why Brian didn’t include backlink count in his own analysis: It doesn’t make for the most shareable of graphs.

The average number of backlinks to all results was 22,771. This is for the page ranking and not the domain as a whole.

As we can see, my data shows very little correlation between backlinks and rankings.

The simple reasoning here is: Not all links are created equally. Ten links from quality, relevant websites have a much greater impact than one thousand links from the same domain.

Referring Domains

On the topic of receiving links from varied domains, I predicted that comparing the number of referring domains to Google rankings should result in data that’s a little more conclusive.

referring

The average number of referring domains to all results was 236.

While I again admit my sample size is small, this data matches pretty much everything else out there I’ve found in regards to the correlation of referring domains and search engine rankings. It basically shows that if you can get a lot of different websites to link to you, that’s going to result in higher rankings (for the most part).

Of course, there is the caveat that ranking highly gives you the chance of more webmasters linking to you, but let me just have my moment here with my first decent chart, OK? 😉

Social Shares

I didn’t expect too much with this one but I had the data so simply decided to chart it.

social-shares

The average number of social shares for all results was 3,823. Again, this was for the page ranking and not the domain as a whole.

The main reason I didn’t expect much from this graph – even if it showed a trending line – is because you can’t distinguish correlation and causation. You can’t show whether social shares helped a website to rank or whether they’re simply a byproduct of writing great content which would have attracted links anyway.

Domain Rank

Domain rating is a metric from Ahrefs which, according to them, “has the highest correlation with the Google search rankings. That’s why I always recommend that Ahrefs Domain Rank be the first SEO metric tool to check whenever you’re analyzing a website.

domain-rating

The average Domain Rank for all results was 63.

I added a trend line to the graph to show that there really wasn’t much change here at all. In fact, Domain Rank was almost perfectly flat across the results.

I imagine if I were monitoring far more ranking positions for each search result then we would see a trend, but there’s nothing out of the ordinary here from page one.

URL Rank

Similar to Domain Rank, Ahrefs also gives a URL Rank rating to specific pages on a website. The majority of results in my testing were internal pages and not homepages, which makes looking at URL Rank (UR) more interesting to me.

url-rating

The average URL Rank for all results was 23.

The results here are certainly a little bit more conclusive. A higher UR seems to have a good correlation with how well a page will rank in Google search results.

Word Count

There have been numerous tests to see whether longer content ranks better in Google so thanks to Word Checker I was able to run these numbers as well.

word-count

The average word count on all results was 1,762.

Again, the argument of correlation versus causation is relevant here. Are pages ranking because they have more words in them or because content with more words in it is likely to attract more links?

Personally, I argue for the latter. I’m far more likely to get links to an in-depth content piece I write rather than something short and sweet. That’s a trend I’ve seen on hundreds of other websites as well.

Behind the Scenes: The Link Building Tactics That Still Work Today

I decided to do put together this report on the state of link building as I’m a little tired of the same SEO advice being rehashed over and over. The thing about our industry is that anyone can start a blog, simply regurgitate what others have said and then instantly appear to be an expert on the topic.

I really like how Aaron Wall of SEO Book put it,

Most of the info created about SEO today is derivative (people who write about SEO but don’t practice it) or people overstating the risks and claiming x and y and z don’t work, can’t work, and will never work.

And then there are people who read an old blog post about how things were x years ago and write as though everything is still the same.

Since I started ViperChill 11 years ago I’ve been testing almost every theory I can when it comes to search engine rankings.

For example, I recently sent 1,000+ clicks to various search results (from around the world) to see if an increased click-through rate (CTR) would influence search engine rankings. Sadly my data didn’t show any noteworthy changes:

clicks

It cost me a few hundred dollars to perform this test and would have made a great blog post if there were any big shifts, but sadly I don’t have any data to support that idea.

I’m always testing, but there isn’t always something to say about my findings.

A Note Before We Get Into ‘Outing’

As I have mentioned in a number of previous blog posts, I will never reveal URL’s when looking at the backlinking strategy of small brands. My experience tells me that big brands will never be affected by my writing and I have proved that on a number of occasions.

I’m about to discuss the slightly shady SEO practices of both Houzz.com and Desk.com, companies both worth billions of dollars (Desk is part of Salesforce). I have dedicated entire blog posts to both of these companies before and there were no repercussions, hence I believe there is zero chance of them having any issues buried deep in a blog post like this one.

As I’ll mention in more detail later, I’ve seen that big brands can “get away” with shadier tactics as long as their overall link profile is natural (and abundant).

Billion Dollar Houzz Prove Widget Links Still Work

In April of 2014 I wrote a blog post about Houzz, the multi-billion dollar home design community.

To summarise much longer commentary, I revealed that Houzz were using their widget to unsuspectingly embed dozens of hard-coded links in the websites of those who used it. Their search traffic grew at a phenomenal rate thanks to the tactic.

Within 24 hours of my blog post about Houzz’s shady tactics going live, they removed all links in their widgets, as shown below.

Unfortunately I do not have a larger graphic for this (it was over two years ago that they had this design) but my prior research provides many additional screenshots.

The problem is that the links they embedded on webmaster websites were hard-coded so even when Houzz changed the widget, those links didn’t disappear and they still benefited from tens of thousands of links from thousands of referring domains.

As you can imagine, their search traffic at the time was through the roof.

Clearly someone from their team read my article and as stated, the hard coded links were removed in less than 24 hours of it going live.

Sadly, Houzz have (partially) gone back to their old ways.

As we can see, Houzz recently added back a link to /photos/ on every single widget their members install on their websites.

houzz-widgets

As per Google’s guidelines, widget links embedded in this way should definitely be no-followed.

Linking to their /photos/ page is smart as it’s essentially a sitemap to the rest of their website, funneling the “link juice” to other strong pages.

Thanks to SEMRush we can see that 7 out of the 10 most high-volume search terms sending traffic to Houzz are actually photos pages.

7outta10

I am aware that widget links are not the only reason why Houzz are ranking for these terms but the whole thing is a little bizarre to me.

The three main things I don’t understand are:

  1. They already retracted after being caught before. Why do it again?
  2. Do they really not care about their users that they can’t put a no-follow on the widget?
  3. They are Houzz. They’re still going to get a ton of search traffic anyway.

The last point is the main one for me. It’s not like they’re some newcomer to the online design space and need to implement these sneaky tactics in order to rank higher.

They’re worth billions of dollars and are expected to IPO next year. Let’s see if the Houzz SEO team are still subscribed to ViperChill. I’ll update this post if there are any changes.

Footer Links Still Work

We already know this from my report on the 16 companies dominating Google in regards to owning a powerful network, but there’s sadly more to the story than that. Big media publishers are not the only ones who get away with putting footer links wherever they can.

In 2013 I wrote an article about how to get a link from SoundCloud.com. The answer today is still the same as it was back then: Give them some software to publicly use on their site and put a footer link back to your website.

Salesforce’s Desk.com continue to do exactly that.

Here’s the footer for SoundCloud (http://help.soundcloud.com)

soundcloud

Here’s the footer for JWPlayer (http://support.jwplayer.com)

jw

Here’s the footer for Wunderlist (http://support.wunderlist.com)

wunderlinst

The list goes on. There are over 1,000 unique websites sending links back to Desk.com with this exact anchor text.

Of course, we don’t have to guess who’s ranking first in Google.

customer-service-software
Note: I removed the ads for a “cleaner” screenshot

This adds further weight to my theory is that as long as you have enough backlinks, you can ignore most of the Google guidelines and still be totally fine.

Marie Haynes has a great article about what is and isn’t “allowed” when it comes to footer links but this tactic certainly toes a very fine line.

Past Link Building Still Holds Strong Today

Even if you aren’t active on online dating sites, you’ve likely heard of Match.com, Tinder and OKCupid.

But what about Mingle2?

It claims 12 million users and is second in Google for ‘Free Online Dating’ yet you’ve probably never heard about it in any form of media.

In fact, you’re more likely to have heard about The Oatmeal.

That’s not a random connection. Matthew Inman started his internet career at SEO company Moz (named SEOmoz at the time) then went on to build the dating site in just 66 hours. His massive success in promoting the platform with viral content and quizzes would later see him sell Mingle2 to Just Say Hi. You probably know he continued to use his amazing talent for creating viral content at The Oatmeal.

For those who aren’t reading a line of text in this post, allow me to put that in graphic form for you:

mingle2

Within a few short months of Matt creating Mingle2 it quickly rose to the top of Google for some very popular search terms. Today, 10 years later, the creative links he built are no doubt helping to sustain those rankings.

mingle2

I don’t want to give too much analysis on this result because I actually think it’s one to watch for how creative Matt was in getting backlinks.

In fact I think if you have some spare time today you should go and analyse their backlinks in more detail. Matthew perfected the art of getting people to want to talk about his content.

As far as link building goes, let’s just say that what they were doing back then would definitely result in a brand being outed today. Those broken guidelines allow Mingle2 to keep their amazing search traffic.

11.7M Reasons Writing Good Content Still Works

For a few years now I’ve considered Steve Kamb (of Nerd Fitness fame) a good friend of mine. That may have something to do with how many Jaegermeister shots we drank together in Cape Town.

I knew Steve was receiving a lot of traffic from Google for his guide to the Paleo diet so I reached out to him to see if he would share any specifics. Especially since the blog post received links from over 800 domains.

Here’s what Steve said,

I wrote the article in Sep 2010 it looks like. In March 0f 2012, Google started to love us all of a sudden sending 76k views. April reached 100,000+ and then it slowly climbed up to a peak in June of 2014 where it was viewed 555,000 times.

Then Google must have changed something and it dropped all the way down over next 6 months to 100k-ish in Dec 2014, where we’ve kind of stabilized over the past 18 months. The pageview count for May 2016 was 87,000.

Steve kindly shared the following graph as proof.

steve

You can click on the picture to view it larger

Even though the article is six years old and has dropped down a few places in Google search results, it still picks up links to this day. Getting real, “earned” links to quality content is far from a dead opportunity.

There are four core reasons I believe Steve’s article still regularly attracts links:

  • Reason #1: He already ranks highly in Google for the term so when people want to link to a guide about the Paleo diet, they see what is ranking and link out
  • Reason #2: Steve wrote one of the best articles on the topic. People simply wouldn’t be linking to it naturally if it wasn’t an incredible resource
  • Reason #3: The article is linked to in the sidebar of every page of his website thus sending it more pageviews than it would have otherwise received (especially since it was written so long ago)
  • Reason #4: He has built a loyal audience of people who genuinely love his content and want to support him in any way they can. It makes sense to them that when they write about the topic, they link to Steve.
  • As you can see, I believe that earning links to your content is far more than just “writing something worth linking to”. It helps if you’ve already built a trusted audience – or you’re willing to work to build one – who would love for more people to hear about your work.

    You Can Buy Your Way Onto Forbes or the Huffington Post

    I get pitched a lot of SEO services via ViperChill and I also actively seek them out myself. It’s always good to know what others are doing since SEO is pretty much my life (as sad as that may sound).

    I regularly receive emails like the one below, revealing specific websites which I can purchase links from.

    paid
    Click the image to view larger

    While the sites in the screenshot above are relatively “small time”, they’re often very safe to use as they have a harder chance of being detected (they’re not going to appear on the first page of an Ahrefs backlink report, for example).

    It’s not just small unknown websites that you can buy your way onto though. I personally know of three different companies who offer the chance to get you on the likes of the Huffington Post for a modest fee.

    opps
    Click the image to view larger

    Once you spend a bit of time reading the various sites you can easily see which websites are selling links and which authors are the most commonly writing them.

    For instance, this article and this article on Business Insider seem to be clearly paid for. There are many others posted by the same author but there are no mentions of the content being sponsored. The author seems overly intent to give credit back to the person he’s writing about (my emails to him did not receive a response so I can’t say for certain).

    I will not link to it but another recent article on Business Insider was heavily promoting a webinar that the featured marketer publicly admitted paying for. The article literally linked straight to his webinars where he was selling a coaching program so the traffic must have made him a few sales.

    Easy-to-Obtain Links Can Dominate the Right Industries

    By easy to obtain I literally mean links that anyone can go out and build for themselves right now with little to no wait time or approval process.

    These sites generally allow you to just sign-up, create some content, and link away. You can see the types of links I’m talking about in the screenshot below.

    easy-lnks

    While these types of links are barely better than no-followed comments on a Japanese cat blog, they can work in the right industries. Particularly industries that are new or would appeal to a certain demographic you never consider i.e. grandmothers who knit or guides on the world’s best rollercoasters.

    Although it may seem like most popular searches are dominated by people who know a thing or two about SEO, it’s certainly not always the case. The site with the links above is relying on Google for almost all of their traffic.

    traffc

    While these certainly aren’t the types of links I recommend building for your own site – unless you need some quick diversity links – it goes to show that they can still work if you’re targeting the right niche.

    Manual Link Hunting Gets Search Traffic

    When you’re averaging 805 links per domain linking to you, there’s probably something a little fishy with your backlink profile. That’s certainly the case with this .info website (I highlighted it’s .info so you can see it’s the same site in both graphics).

    refer

    If my yellow highlighter effect is doing its job then you should be impressed to see that they’re ranking for over 1 million keywords in search results.

    And these are not just any old keywords either. Some of them are being searched for tens of thousands of times per month. You can see the same website displayed in the SEMRush graph below.

    semrush

    Without adding yet another screenshot to this already image-heavy article, just trust me when I say the quality of their backlinks leaves a lot to be desired.

    That being said, I do have some respect for the owner in how much work they’ve done to build these links. They ‘abused’ a few opportunities but put the work in. The links come from sources like:

    • Wikipedia external links
    • Comments on news articles (via actual news websites) with relevant stories
    • Opportunities likely found via links their competitors earned

    It actually gave me a great idea for a niche to get involved in as well, so although the work for this blog post was immense, I’ve found a number of opportunities because of it.

    Private Blog Networks Still Work Very Well

    Being totally honest, I expected to find more link networks in my research than I did. Especially because I was monitoring the type of industries where this practice is likely to be more common.

    Here is how an obvious network looks when you analyse their backlinks:

    network

    The example above is actually what I would call a “good” example. Meaning that they websites ranking and linking to each other are actually good sites and far more searcher-friendly than the typical blog network I am sure you can picture in your mind.

    It simply provided a nice screenshot to illustrate my point into how these networks work.

    Of course we’ve already learned that if you have thousands of links pointing to a number of websites you own, you can interlink them and dominate Google search results.

    Update: Some commenters seem angry that I “only” found as many PBN’s as I did.

    Two things to note: I found more than are in the chart above, I just didn’t rate them as being the biggest contributing factor in why a website ranks.

    The number could also be lower because of people hiding their networks from Ahrefs. I may do a smaller version of this study again with something like Link Research Tools or Monitor Backlinks (I’ll have to check if they use their own network) which people are less likely to block.

    You Can Get Dozens of .EDU Backlinks for $1,000

    Ever since I started SEO at 16 years old and spent countless hours browsing the Web Workshop forums (no longer online) I’ve heard about the power of .edu (education / university) backlinks.

    It makes sense that these links would pass a lot of authority because of the sites they’re coming from. They’re certainly not easy to attain naturally: ViperChill has over 100,000 links yet only 8 of them are from .edu sites.

    I don’t even know how I received those 8, since the university of Calgary link to a blog post I wrote five years ago which no longer exists and Australia’s Newcastle University is somehow linking to me via pingbacks.

    One tactic that I’ve found is becoming increasingly common in order to obtain .edu links is to offer a ‘scholarship’ on your website and receive dozens if not hundreds of .edu links to your site in return.

    It’s certainly not new by any means. With a bit of sleuthing around you can see sites – clearly just offering a scholarship for a link – have been employing this for a number of years.

    launch

    I’ve tried my best to be respectful to the site owner and not reveal their website but anyone who is using this tactic and thinks they’re doing so under the radar really has no idea what that phrase means.

    As this is a behind the scenes report on what still works in 2016, I wanted to make it clear that this is still happening today and people are benefiting from it massively.

    scholar

    There will obviously be people who are pissed off I have “exposed” this tactic but to me there is nothing shadier then making students believe there’s a chance they could save money on their education yet they probably have no chance to do so at all.

    Is there really anyone checking to see if a coupon website launched in 2016 is going to keep to their $3,000 promise?

    Exactly.

    While these guys are supposedly offering $3,000, I’ve found some offering as little as $1,000 and still picking up a large number of links.

    The Future of Link Building

    It would be wrong of me to write a huge report on link building without speculating on what the future of link building might entail. We all know that backlinks are a large part of why websites rank today, but will that still be the case tomorrow?

    The SEO industry is fortunate to have enough bright minds that people tackle problems like this. My good friend Jon from PointBlankSEO wrote a great report to try and answer this very question.

    In his conclusion, Jon makes an excellent point:

    The real threat is more foundational than links. Justin Briggs explained it best in his response earlier. The aspect of ranking a page organically in Google’s results has slowly declined in value, both because of other SERP features & search ads. There’s still a ton of money to be made, but we should work like we’re living on borrowed time.

    Today, natural organic search results are lower down in listings than ever before.

    Mobile results are spaced further apart. “Map packs” in search results take up half of the screen on a desktop. Google’s one box tries to answer user queries straight from the results page.

    I don’t see any major ranking competitor to links in the near future. The entire Google algorithm which provided better results than Altavista and Yahoo back in the day was built on links and 18 years later they’re still a key factor in why websites rank.

    That being said, my main concern is where SEO will be in three to five years rather than what matters to rank. We’ll always figure out the last part. The first part is out of our control.

    Teaser: There’s One Tactic That is Dominating Them All

    Over the past 18 months I’ve found one link building tactic to be working incredibly well. It’s not brand new in the sense of “Wow, University’s give out links so easily” but in the sense of “here’s how to make all of these current options work even better.”

    If I wrote about it, I would probably lose a large chunk of my audience, but that’s something I’m willing to do.

    Next week I’m going to introduce PIN’s, a new way to conduct link building which could fit anywhere on the spectrum of whitehat to blackhat.

    It’s a very risky topic to cover so for that reason I want to dedicate an entire article to it, rather than just add another section to this report which could be taken entirely out of context.

    If you’re new to ViperChill, enter your email in the box below (or in the right sidebar) to make sure you don’t miss that update. I’ll send it out the minute it goes live.

    Thank you so much, as always, for reading.


Enjoy this post? Please leave a comment below…

Comments are my number one indicator as to which posts people enjoyed the most, so your feedback really does help me. If you have any questions, feel free to ask those as well…



Source link

The Most Power Comprehensive Lead Generate Toolkits
Click Banner

Leave a Reply

Your email address will not be published. Required fields are marked *

/* */
the-state-of-link-building-2016-what-i-learned-manually-analysing-1-000-search-results